fix(spans): Avoid back pressure in the segments consumer #88419
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
untitaker
approved these changes
Apr 1, 2025
andrewshie-sentry
pushed a commit
that referenced
this pull request
Apr 8, 2025
In multi processing mode, the segments consumer continuously triggered backpressure. This is because we batch segments in the multiprocessing step. When a batch of segments is flushed by a process, the pipeline unfolds the spans, which immediately exhausts the produce buffer. For example, at a batch size of 100 with 4 processes and assuming an average of 40 spans per segment, we write up to 16k spans if all processes finish around the same time. This already exceeds the default buffer size of 10k. To avoid manual tuning, we've run long-term tests and determined that the p95 spans per segment is at 350. We use this along with the number of processes and the batch size to compute a safer upper bound for the produce buffer. The consumer is primarily CPU and I/O bound, so we can afford the additional memory usage.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
In multi processing mode, the segments consumer continuously triggered
backpressure. This is because we batch segments in the multiprocessing
step. When a batch of segments is flushed by a process, the pipeline
unfolds the spans, which immediately exhausts the produce buffer.
For example, at a batch size of 100 with 4 processes and assuming an
average of 40 spans per segment, we write up to 16k spans if all
processes finish around the same time. This already exceeds the default
buffer size of 10k.
To avoid manual tuning, we've run long-term tests and determined that
the p95 spans per segment is at 350. We use this along with the number
of processes and the batch size to compute a safer upper bound for the
produce buffer. The consumer is primarily CPU and I/O bound, so we can
afford the additional memory usage.